Comparison of clinical geneticist and computer visual attention in assessing genetic conditions.

Artificial intelligence (AI) for facial diagnostics is increasingly used in the genetics clinic to evaluate patients with potential genetic conditions. Current approaches focus on one type of AI called Deep Learning (DL). While DL- based facial diagnostic platforms have a high accuracy rate for many conditions, less is understood about how this technology assesses and classifies (categorizes) images, and how this compares to humans. To compare human and computer attention, we performed eye-tracking analyses of geneticist clinicians (n = 22) and non-clinicians (n = 22) who viewed images of people with 10 different genetic conditions, as well as images of unaffected individuals. We calculated the Intersection-over-Union (IoU) and Kullback-Leibler divergence (KL) to compare the visual attentions of the two participant groups, and then the clinician group against the saliency maps of our deep learning classifier. We found that human visual attention differs greatly from DL model's saliency results. Averaging over all the test images, IoU and KL metric for the successful (accurate) clinician visual attentions versus the saliency maps were 0.15 and 11.15, respectively. Individuals also tend to have a specific pattern of image inspection, and clinicians demonstrate different visual attention patterns than non-clinicians (IoU and KL of clinicians versus non-clinicians were 0.47 and 2.73, respectively). This study shows that humans (at different levels of expertise) and a computer vision model examine images differently. Understanding these differences can improve the design and use of AI tools, and lead to more meaningful interactions between clinicians and AI technologies.

Approximating facial expression effects on diagnostic accuracy via generative AI in medical genetics.

Artificial intelligence (AI) is increasingly used in genomics research and practice, and generative AI has garnered significant recent attention. In clinical applications of generative AI, aspects of the underlying datasets can impact results, and confounders should be studied and mitigated. One example involves the facial expressions of people with genetic conditions. Stereotypically, Williams (WS) and Angelman (AS) syndromes are associated with a "happy" demeanor, including a smiling expression. Clinical geneticists may be more likely to identify these conditions in images of smiling individuals. To study the impact of facial expression, we analyzed publicly available facial images of approximately 3500 individuals with genetic conditions. Using a deep learning (DL) image classifier, we found that WS and AS images with non-smiling expressions had significantly lower prediction probabilities for the correct syndrome labels than those with smiling expressions. This was not seen for 22q11.2 deletion and Noonan syndromes, which are not associated with a smiling expression. To further explore the effect of facial expressions, we computationally altered the facial expressions for these images. We trained HyperStyle, a GAN-inversion technique compatible with StyleGAN2, to determine the vector representations of our images. Then, following the concept of InterfaceGAN, we edited these vectors to recreate the original images in a phenotypically accurate way but with a different facial expression. Through online surveys and an eye-tracking experiment, we examined how altered facial expressions affect the performance of human experts. We overall found that facial expression is associated with diagnostic accuracy variably in different genetic conditions.

The effect of GLP-1 receptor agonist use on negative evaluations of women with higher and lower body weight.

BACKGROUND: GLP-1 receptor agonists (GLP-1 RAs) have dramatically altered obesity treatment. Media reports suggest that GLP-1 RAs users often report feeling judged for taking a "shortcut" to lose weight, which may be related to negative stereotypes toward people with larger bodies. Media reports also describe negative attitudes about lean people who take GLP-1 RAs to enhance their appearance. The present research used a 2 × 2 experimental design to test the effects of GLP-1 RA use and body size on attitudes and egocentric impressions. SUBJECTS/METHODS: A sample of 357 U.S. adults (Mage = 37.8, SD = 13) were randomly assigned to read about a woman, who either was lean or had obesity, and who lost 15% of her body weight either with diet/exercise or a GLP-1 RA. Participants answered questions measuring endorsement of negative weight-related stereotypes and egocentric attitudes toward the woman, as well as beliefs that she took a shortcut to lose weight and beliefs that biogenetic factors caused her baseline weight. RESULTS: Negative evaluations and egocentric impressions were stronger toward a woman who lost weight with a GLP-1 RA compared to diet/exercise. Losing weight with a GLP-1 RA led to stronger negative evaluations through higher weight loss shortcut beliefs irrespective of body size. Losing weight with a GLP-1 RA also led to higher egocentric impressions through higher shortcut beliefs, and this effect was stronger for a lean woman. Finally, losing weight with a GLP-1 RA led to more negative evaluations through stronger endorsement of biogenetic causal beliefs for a lean woman only. CONCLUSIONS: This timely study provides evidence that people with larger and smaller bodies alike are at-risk for being judged for using GLP-1 RAs due to beliefs that these medications are a shortcut. Findings also demonstrate novel reactions related to egotism when weight loss is achieved with pharmacological interventions. PRE-REGISTRATION AND DATA: osf.io/xme4w.

Measuring the therapeutic bond in genetic counseling: Testing measurement error in the bond subscale of the Working Alliance Inventory.

The therapeutic relationship is a key component of successful genetic counseling. In psychotherapy, a strong therapeutic relationship can improve patient health outcomes and a poor relationship can worsen psychological functioning. Investigation of the therapeutic relationship in genetic counseling has shown evidence for a similar pattern. Reliable measurement of the therapeutic relationship is necessary for consistency across studies in the genetic counseling context. One measure that has been adapted for use in genetic counseling is the Working Alliance Inventory (WAI). However, there have been no studies of the factor structure or item-level method bias analyses for the genetic counseling-adapted version of the WAI. The goal of this study was to test the factor structure of the WAI observer version (WAI-O) bond subscale and assess method bias in a genetic counseling context. We hypothesized that differences in factor structures would exist for items that were positively (n = 9) versus negatively (n = 3) worded (reverse coded). Secondary data analysis was performed on two data sets that utilized the WAI-O in genetic counseling contexts. Data set 1 used simulated genetic counseling sessions that were judged by analog clients recruited through crowdsourcing platforms (N = 861). Data set 2 was conducted with genetic counseling clients, and sessions were evaluated by a research team (N = 120). Principal axis factor analysis with oblique oblimin rotation supported a two-factor solution for the WAI-O bond subscale across data sets. Items factored based on wording, with the positively worded items loading together and the negatively worded items loading on the second factor. Confirmatory factor analyses supported the removal of all negatively worded items from the instrument across data sets. Results suggest that the negatively worded items on the WAI-O may be capturing a construct inconsistent with the positively worded items and support rewording and/or excluding them from use for a more reliable measure of the therapeutic bond.

Medical Extended Reality Trials: Building Robust Comparators, Controls, and Sham.

The explosive pace of development and research in medical extended reality (MXR) is a testament to its promise for health care and medicine. In comparison with this growth, there is a relative sparsity of rigorous clinical trials that establish the efficacy and effectiveness of these interventions. Explicating mechanisms of action across clinical areas and MXR applications is another major area of need. A primary impediment to these goals is a lack of frameworks for trial design, more specifically, the selection of appropriate controls that effectively address unique elements of MXR. This paper delineates a framework for designing controls, sham conditions, and comparators, as well as proposed considerations for MXR trial designs. Special consideration is given to the design of sham conditions. Improved designs would enable more robust findings and the development of generalizable knowledge that could be adopted across MXR interventions.

Characterizing Precision Nutrition Discourse on Twitter: Quantitative Content Analysis.

BACKGROUND: It is possible that tailoring dietary approaches to an individual's genomic profile could provide optimal dietary inputs for biological functioning and support adherence to dietary management protocols. The science required for such nutrigenetic and nutrigenomic profiling is not yet considered ready for broad application by the scientific and medical communities; however, many personalized nutrition products are available in the marketplace, creating the potential for hype and misleading information on social media. Twitter provides a unique big data source that provides real-time information. Therefore, it has the potential to disseminate evidence-based health information, as well as misinformation. OBJECTIVE: We sought to characterize the landscape of precision nutrition content on Twitter, with a specific focus on nutrigenetics and nutrigenomics. We focused on tweet authors, types of content, and presence of misinformation. METHODS: Twitter Archiver was used to capture tweets from September 1, 2020, to December 1, 2020, using keywords related to nutrition and genetics. A random sample of tweets was coded using quantitative content analysis by 4 trained coders. Codebook-driven, quantified information about tweet authors, content details, information quality, and engagement metrics were compiled and analyzed. RESULTS: The most common categories of tweets were precision nutrition products and nutrigenomic concepts. About a quarter (132/504, 26.2%) of tweet authors presented themselves as science experts, medicine experts, or both. Nutrigenetics concepts most frequently came from authors with science and medicine expertise, and tweets about the influence of genes on weight were more likely to come from authors with neither type of expertise. A total of 14.9% (75/504) of the tweets were noted to contain untrue information; these were most likely to occur in the nutrigenomics concepts topic category. CONCLUSIONS: By evaluating social media discourse on precision nutrition on Twitter, we made several observations about the content available in the information environment through which individuals can learn about related concepts and products. Tweet content was consistent with the indicators of medical hype, and the inclusion of potentially misleading and untrue information was common. We identified a contingent of users with scientific and medical expertise who were active in discussing nutrigenomics concepts and products and who may be encouraged to share credible expert advice on precision nutrition and tackle false information as this technology develops.

Educational considerations based on medical student use of polygenic risk information and apparent race in a simulated consultation.

PURPOSE: To craft evidence-based educational approaches related to polygenic risk score (PRS) implementation, it is crucial to forecast issues and biases that may arise when PRS are introduced in clinical care. METHODS: Medical students (N = 84) were randomized to a simulated primary care encounter with a Black or White virtual reality-based patient and received either a direct-to-consumer-style PRS report for 5 common complex conditions or control information. The virtual patient inquired about 2 health concerns and her genetic report in the encounter. Data sources included participants' verbalizations in the simulation, care plan recommendations, and self-report outcomes. RESULTS: When medical students received PRSs, they rated the patient as less healthy and requiring more strict advice. Patterns suggest that PRSs influenced specific medical recommendations related to the patient's concerns, despite student reports that participants did not use it for that purpose. We observed complex patterns regarding the effect of patient race on recommendations and behaviors. CONCLUSION: Educational approaches should consider potential unintentional influences of PRSs on decision-making and evaluate ways that they may be applied inconsistently across patients from different racial groups.

Exploring the impact of genetic beliefs about specific eating behaviors on dietary self-efficacy.

Future personalized approaches to weight management are likely to include consideration of genetic influences on eating behaviors. This study explores whether genetic beliefs about eating behaviors influence dietary self-efficacy and confidence. In a survey of 261 individuals of various weight statuses, we find that endorsing genetic causes of two specific eating behaviors (taste preference and disinhibition) predicts poorer dietary self-efficacy for people who exhibit these eating behaviors. This suggests there may be utility to considering eating behaviors individually when it comes to predicting the influence of genetic information provision in the service of precision medicine interventions. Individuals with high disinhibited eating and/or bitter taster status may be particularly sensitive to interpreting genetic predisposition information in ways that undercut self-efficacy and confidence.

Cybersickness Variability by Race: Findings From 6 Studies and a Mini Meta-analysis.

BACKGROUND: With the influx of medical virtual reality (VR) technologies, cybersickness has transitioned from a nuisance experienced during leisure activities to a potential safety and efficacy concern for patients and clinicians. To improve health equity, it is important to understand any potential differences in cybersickness propensity among demographic groups, including racial groups. OBJECTIVE: This study aims to explore whether cybersickness propensity differs across racial groups. METHODS: We collected self-reported cybersickness ratings from 6 racially diverse independent samples within 1 laboratory group (N=931). In these studies, the participants were asked to perform tasks in VR such as traversing environments, pointing at and selecting objects, and interacting with virtual humans. RESULTS: Significant racial differences in cybersickness were found in 50% (3/6) of studies. A mini meta-analysis revealed that, on average, Black participants reported approximately one-third of SD less cybersickness than White participants (Cohen d=-0.31; P<.001), regardless of the nature of the VR experience. There was no overall difference in reported cybersickness between the Asian and White participants (Cohen d=-0.11; P=.51). CONCLUSIONS: Racial differences in cybersickness indicate that researchers, practitioners, and regulators should consider patient demographics when evaluating VR health intervention outcomes. These findings lay the groundwork for future studies that may explore racial differences in cybersickness directly.

Investigating the Efficacy of Genetic, Environmental, and Multifactorial Risk Information When Communicating Obesity Risk to Parents of Young Children.

BACKGROUND: Effectively communicating with parents about children's obesity risk is of critical importance for preventive medicine and public health. PURPOSE: The current study investigates the efficacy of communications focused on two primary causes of obesity: genes and environment. METHODS: We compared parental feeding responses to messages focused on (i) genetics alone, (ii) family environment alone, (iii) genetics-family environment interaction (G × FE), and (iv) no causal message. We also examined whether parental guilt mediates the effect of message type on feeding. Our sample consisted of 190 parents, half mothers and half fathers, of children 3-7 years old. After receiving one of the four types of messages, parents chose foods for their child using the Virtual Reality Buffet measure. Parents responded to questionnaires in the lab and at 1-week follow-up. RESULTS: In the VR Buffet, parents did not feed their children differently in message provision conditions versus control. There were, however, differences among message provision conditions wherein mothers who received any genetic information chose higher-calorie meals in the VR Buffet. At 1-week follow-up, parents who received information about genetics alone reported feeding their child more junk food and fatty meat on self-report food frequency assessments; there were no such differences for sugary beverages, sugary foods, or fast foods. Parental guilt was typically higher for participants who received family environment information alone but did not mediate the relation between information provision and feeding outcomes. CONCLUSIONS: While none of the messages improved feeding above the control condition, GxFE messages were associated with a better overall profile of outcomes. As such, it may be beneficial for messaging for parents about children's obesity risk to include content that reflects the complexity of genetic and environmental contributions to obesity risk.

Parental reward-based eating drive predicts parents' feeding behaviors and Children's ultra-processed food intake.

Reward-based eating drive is associated with individual consumption, but there has been a paucity of research on the relationships between parental reward-based eating, child feeding behaviors, and child food consumption. Child feeding behaviors likely to be associated with parental reward-based eating drive include the provision of ultra-processed foods, as they are designed to be hyperpalatable and are associated with disordered food intake. The present study uses a virtual reality (VR) buffet restaurant environment to examine parents' food choice behaviors for their children and a food frequency assessment to measure the children's reported consumption over the course of a week. Results found that parental reward-based eating drive significantly predicted ultra-processed calories chosen by parents for their children in the VR Buffet, as well as the amount of ultra-processed food children ate according to the food frequency assessment. Both of these effects were significantly mediated by the healthfulness of the home food environment. This study is among the first to demonstrate associations between parental reward-based eating drive and child-focused food behavior and to elucidate a mediating effect of the home food environment on such relationships. These findings may be useful for the development of family-based interventions to improve child feeding and ultimately child health.

Parents' genetic attributions for children's eating behaviors: Relationships with beliefs, emotions, and food choice behavior.

Considering genetic influences on children's eating behavior could result in reduced self-efficacy for healthy child feeding and less healthy feeding behavior among parents. Indeed, one's eating behaviors are typically thought of as the volitional aspects of weight management that one can directly control. The current study assessed parental genetic attributions for their child's eating behavior, and relationships between these attributions and self-efficacy, guilt, and feeding behaviors. Participants included 190 parents of a child between 4 and 7 years old. Parents' genetic attributions for child eating behaviors were lower than genetic attributions related to child weight. Higher genetic attributions for child eating behaviors were related to lower self-efficacy for feeding the right amounts of food, higher-calorie food choices for the child in a virtual reality-based buffet simulation, and higher levels of guilt. The current findings suggest that heightened beliefs about role of genetics in children's eating behavior is associated with maladaptive affect and behavior among parents. This should be kept in mind when considering whether, when, and how to provide information to parents highlighting the role of genetics in children's eating behavior.

A Virtual Home for the Virtual Clinical Trial.

Virtual clinical trials (VCTs) can satisfy the need for rigorous clinical trials by using distributed technological solutions that eliminate the need for a physical trial site. This report explores potential benefits of using virtual reality (VR) to provide a "virtual site" for VCTs, a shared immersive hub in which VCT participants could experience elements of the trial and interact with the trial team. VR is a communication technology that has been emerging alongside the development of VCTs, although they have never been merged in a substantial way. Many of the gaps within the VCT paradigm are areas in which VR excels. VR environments are standardized and precisely uniform, the technology allows introduction of an almost endless set of stimuli to participants' visual and auditory systems, and VR systems are adept at capturing precise movement and behavioral data. Although VR has not yet found its way into VCTs, much of the groundwork for such integration has been laid through research and technological development achieved in the past few years. Future implementation of VR within VCTs could move us from site-less trials to those with a virtual site serving as a hub for trial information provision, interaction with trial representatives, administration of evaluations and assessments, and more.

Designing and Assessing Multilevel Interventions to Improve Minority Health and Reduce Health Disparities.

Multilevel interventions can be uniquely effective at addressing minority health and health disparities, but they pose substantial methodological, data analytic, and assessment challenges that must be considered when designing and applying interventions and assessment. To facilitate the adoption of multilevel interventions to reduce health disparities, we outline areas of need in filling existing operational challenges to the design and assessment of multilevel interventions. We discuss areas of development that address overarching constructs inherent in multilevel interventions, with a particular focus on their application to minority health and health disparities. Our approach will prove useful to researchers, as it allows them to integrate information related to health disparities research into the framework of broader constructs with which they are familiar. We urge researchers to prioritize building transdisciplinary teams and the skills needed to overcome the challenges in designing and assessing multilevel interventions, as even small contributions can accelerate progress toward improving minority health and reducing health disparities. To make substantial progress, however, a concerted and strategic effort, including work to advance analytic techniques and measures, is needed.

Effects of Fruit and Vegetable Feeding Messages on Mothers and Fathers: Interactions Between Emotional State and Health Message Framing.

BACKGROUND: There is a pressing need to craft optimal public health messages promoting healthy feeding behaviors among parents. How these messages influence such feeding decisions are affected by multiple interactive factors including emotional states, message framing, and gender, but these factors have not been studied in the domain of parents' feeding of their children. PURPOSE: To evaluate the role of message framing, emotional state, and parent gender on feeding choices that parents make for their children. METHODS: In 2016-2017, 190 parents (126 mothers) of 4- to 7-year-old children were randomly assigned to an anger or fear emotion induction and read either a gain- or loss-framed message about the importance of children's fruit and vegetable (FV) consumption prior to choosing foods for their child from a virtual reality buffet. RESULTS: Mothers in an angry state who received a gain-framed message chose relatively more FV for their child in the virtual buffet, F(3, 180) = 4.77, p = .027. However, fathers in this group did not feed more FV, but rather reported greater intention to improve future FV feeding, F(3, 180) = 4.91, p = .028. CONCLUSIONS: Providing gain-framed messages to parents, particularly mothers, in an anger state may be most effective for motivating healthy dietary choices for children. CLINICAL TRIAL INFORMATION: clinicaltrials.gov NCT02622035.

Modeling Dynamic Food Choice Processes to Understand Dietary Intervention Effects.

Background: Meal construction is largely governed by nonconscious and habit-based processes that can be represented as a collection of in dividual, micro-level food choices that eventually give rise to a final plate. Despite this, dietary behavior intervention research rarely captures these micro-level food choice processes, instead measuring outcomes at aggregated levels. This is due in part to a dearth of analytic techniques to model these dynamic time-series events. Purpose: The current article addresses this limitation by applying a generalization of the relational event framework to model micro-level food choice behavior following an educational intervention. Method: Relational event modeling was used to model the food choices that 221 mothers made for their child following receipt of an information-based intervention. Participants were randomized to receive either (a) control information; (b) childhood obesity risk information; (c) childhood obesity risk information plus a personalized family history-based risk estimate for their child. Participants then made food choices for their child in a virtual reality-based food buffet simulation. Results: Micro-level aspects of the built environment, such as the ordering of each food in the buffet, were influential. Other dynamic processes such as choice inertia also influenced food selection. Among participants receiving the strongest intervention condition, choice inertia decreased and the overall rate of food selection increased. Conclusions: Modeling food selection processes can elucidate the points at which interventions exert their influence. Researchers can leverage these findings to gain insight into nonconscious and uncontrollable aspects of food selection that influence dietary outcomes, which can ultimately improve the design of dietary interventions.

Validity of assessing child feeding with virtual reality.

Assessment of parents' child feeding behavior is challenging, and there is need for additional methodological approaches. Virtual reality technology allows for the creation of behavioral measures, and its implementation overcomes several limitations of existing methods. This report evaluates the validity and usability of the Virtual Reality (VR) Buffet among a sample of 52 parents of children aged 3-7. Participants served a meal of pasta and apple juice in both a virtual setting and real-world setting (counterbalanced and separated by a distractor task). They then created another meal for their child, this time choosing from the full set of food options in the VR Buffet. Finally, participants completed a food estimation task followed by a questionnaire, which assessed their perceptions of the VR Buffet. Results revealed that the amount of virtual pasta served by parents correlated significantly with the amount of real pasta they served, rs = 0.613, p < .0001, as did served amounts of virtual and real apple juice, rs = 0.822, p < .0001. Furthermore, parents' perception of the calorie content of chosen foods was significantly correlated with observed calorie content (rs = 0.438, p = .002), and parents agreed that they would feed the meal they created to their child (M = 4.43, SD = 0.82 on a 1-5 scale). The data presented here demonstrate that parent behavior in the VR Buffet is highly related to real-world behavior, and that the tool is well-rated by parents. Given the data presented and the potential benefits of the abundant behavioral data the VR Buffet can provide, we conclude that it is a valid and needed addition to the array of tools for assessing feeding behavior.

Identifying eating behavior phenotypes and their correlates: A novel direction toward improving weight management interventions.

Common reports of over-response to food cues, difficulties with calorie restriction, and difficulty adhering to dietary guidelines suggest that eating behaviors could be interrelated in ways that influence weight management efforts. The feasibility of identifying robust eating phenotypes (showing face, content, and criterion validity) was explored based on well-validated individual eating behavior assessments. Adults (n = 260; mean age 34 years) completed online questionnaires with measurements of nine eating behaviors including: appetite for palatable foods, binge eating, bitter taste sensitivity, disinhibition, food neophobia, pickiness and satiety responsiveness. Discovery-based visualization procedures that have the combined strengths of heatmaps and hierarchical clustering were used to investigate: 1) how eating behaviors cluster, 2) how participants can be grouped within eating behavior clusters, and 3) whether group clustering is associated with body mass index (BMI) and dietary self-efficacy levels. Two distinct eating behavior clusters and participant groups that aligned within these clusters were identified: one with higher drive to eat and another with food avoidance behaviors. Participants' BMI (p = 0.0002) and dietary self-efficacy (p < 0.0001) were associated with cluster membership. Eating behavior clusters showed content and criterion validity based on their association with BMI (associated, but not entirely overlapping) and dietary self-efficacy. Identifying eating behavior phenotypes appears viable. These efforts could be expanded and ultimately inform tailored weight management interventions.

Framed, Interactive Theory-Driven Texting: Effects of Message Framing on Health Behavior Change for Weight Loss.

More approaches to support weight control are needed, especially among racial minorities who shoulder a disproportionate obesity burden. Using an approach influenced by regulatory fit theory, we conducted a 28-day, 4-arm experimental trial with 89 obese adults recruited from urban, predominantly African American churches to ascertain the efficacy of framed text messages to motivate behaviors conducive to weight loss. Participants were assigned to receive message framing that was matched versus mismatched to their motivational orientation. Results were mixed overall; however, matched texts elicited greater motivation to change eating and exercise behavior, suggesting promise in using motivational approaches to tailor messages.

"It runs in my family ": The association of perceived family history with body dissatisfaction and weight bias internalization among overweight women.

Aspects of poor body acceptance (BA), such as internalized weight bias and dissatisfaction with one's shape and size, are the strongest predictors of disordered eating and are associated with reduced engagement in healthy behaviors. Perceiving oneself as having a family history of overweight (PFH) could boost BA by increasing attributions for inherited, biological causes of weight. A community sample of 289 women who were overweight from the Washington, DC metropolitan area who were dissatisfied with their current weight (68% Black; 32% White) enrolled in this study in 2012. PFH of overweight was associated with decreased internalized weight bias among white women and marginally increased body shape satisfaction generally. The relationship between PFH and BA was not explained by biological attributions for weight. Perceptions that overweight runs in one's family can be protective with respect to BA. This is suggestive of the potential benefit of integrating family-based approaches into weight management interventions.